IlinValery/labeling_stat

Labling repo statistic GUIMC

VENV setup

$ python3 -m venv env
$ source env/bin/activate
#for deactivate use "deactivate" in command line
$ pip install -r req.txt

Folder setup and run

#move all dirs from 4_Сводка to catalog/

$ sudo chmod +x collect_xml.sh
$ ./collect_xml.sh

#as result - new xml file "data.xml"

$ python get_statistic.py

#result - 3 files:
#"output_total.csv" - count frames in all xmls files
#"output_with_light.csv" - count frames in all xmls files with lights_types
#"summary.xls" - files list with objects and lights

Use Jupyter

$ jupyter notebook #from terminal
#open pandas_analys and relax

Data Augmentation

Here we use Augmentor and Tensorflow.image module instruments for Python3.
For launch help of our Augmetator we need to use

$ python augmentation.py --help

usage: augmentation.py [-h] [--folder FOLDER] [--videoname VIDEONAME]
                       [--additional ADDITIONAL]

optional arguments:
  -h, --help            show this help message and exit
  --folder FOLDER       Folder with *.jpg files. for example: --folder
                        augmentation_data/carrot_L2_002/
  --videoname VIDEONAME
                        Video number for files *.jpg. For example: 004
  --additional ADDITIONAL
                        Count of additional files. System create same count
                        augmented data plus --additional count

For correct launch script use:

$ python augmentation.py --folder augmentation_data/carrot_L2_002/ --videoname 004